National vital statistics in the United States present data in terms of race, sex, and age, treated as biological variables. Some races are clearly of more interest than others: data are usually available for whites and blacks, and increasingly for Hispanics, but seldom for Native Americans or Asians and Pacific Islanders. These data indicate that white men and women generally have the best health and that men and women, within each racial/ethnic group, have different patterns of disease. Obviously, the health status of men and women differs for conditions related to reproduction, but it differs for many nonreproductive conditions as well. In national health data, patterns of disease by race and sex are emphasized while social class differences are ignored. This article discusses how race and sex became such all-important, self-evident categories in 19th and 20th century biomedical thought and practice. It examines the consequences of these categories for knowledge about health and for the provision of health care. It then presents alternative approaches to understanding the relationship between race/ethnicity, gender, and health, with reference to the neglected category of social class.